{"id":"https://openalex.org/W7164837420","doi":"https://doi.org/10.1109/lca.2026.3703423","title":"On the Feasibility Boundaries of LLM Prefill Inference on Unified-Memory Edge SoCs","display_name":"On the Feasibility Boundaries of LLM Prefill Inference on Unified-Memory Edge SoCs","publication_year":2026,"publication_date":"2026-06-15","ids":{"openalex":"https://openalex.org/W7164837420","doi":"https://doi.org/10.1109/lca.2026.3703423"},"language":null,"primary_location":{"id":"doi:10.1109/lca.2026.3703423","is_oa":false,"landing_page_url":"https://doi.org/10.1109/lca.2026.3703423","pdf_url":null,"source":{"id":"https://openalex.org/S17643076","display_name":"IEEE Computer Architecture Letters","issn_l":"1556-6056","issn":["1556-6056","1556-6064","2473-2575"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Computer Architecture Letters","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5138643384","display_name":"Juyeon Lee","orcid":null},"institutions":[{"id":"https://openalex.org/I139264467","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542","country_code":"KR","type":"education","lineage":["https://openalex.org/I139264467"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Juyeon Lee","raw_affiliation_strings":["Computer Architecture and Parallel Processing Lab (CAPP), Department of Electrical and Computer Engineering, Seoul National University, Seoul, South Korea"],"raw_orcid":"https://orcid.org/0009-0008-0947-7370","affiliations":[{"raw_affiliation_string":"Computer Architecture and Parallel Processing Lab (CAPP), Department of Electrical and Computer Engineering, Seoul National University, Seoul, South Korea","institution_ids":["https://openalex.org/I139264467"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101539395","display_name":"Jiayue Wang","orcid":"https://orcid.org/0000-0002-0606-8132"},"institutions":[{"id":"https://openalex.org/I139264467","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542","country_code":"KR","type":"education","lineage":["https://openalex.org/I139264467"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Jiayue Wang","raw_affiliation_strings":["Computer Architecture and Parallel Processing Lab (CAPP), Department of Electrical and Computer Engineering, Seoul National University, Seoul, South Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Computer Architecture and Parallel Processing Lab (CAPP), Department of Electrical and Computer Engineering, Seoul National University, Seoul, South Korea","institution_ids":["https://openalex.org/I139264467"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5109987997","display_name":"Soojung Ryu","orcid":null},"institutions":[{"id":"https://openalex.org/I139264467","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542","country_code":"KR","type":"education","lineage":["https://openalex.org/I139264467"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Soojung Ryu","raw_affiliation_strings":["Computer Architecture and Parallel Processing Lab (CAPP), Department of Electrical and Computer Engineering, Seoul National University, Seoul, South Korea"],"raw_orcid":"https://orcid.org/0009-0000-5228-9801","affiliations":[{"raw_affiliation_string":"Computer Architecture and Parallel Processing Lab (CAPP), Department of Electrical and Computer Engineering, Seoul National University, Seoul, South Korea","institution_ids":["https://openalex.org/I139264467"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5102861073","display_name":"Hyuk-Jae Lee","orcid":"https://orcid.org/0000-0001-8895-9117"},"institutions":[{"id":"https://openalex.org/I139264467","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542","country_code":"KR","type":"education","lineage":["https://openalex.org/I139264467"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Hyuk-Jae Lee","raw_affiliation_strings":["Computer Architecture and Parallel Processing Lab (CAPP), Department of Electrical and Computer Engineering, Seoul National University, Seoul, South Korea"],"raw_orcid":"https://orcid.org/0000-0001-6811-9647","affiliations":[{"raw_affiliation_string":"Computer Architecture and Parallel Processing Lab (CAPP), Department of Electrical and Computer Engineering, Seoul National University, Seoul, South Korea","institution_ids":["https://openalex.org/I139264467"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I139264467"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.80482358,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"25","issue":"2","first_page":"267","last_page":"270"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11032","display_name":"VLSI and Analog Circuit Testing","score":0.24230000376701355,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T11032","display_name":"VLSI and Analog Circuit Testing","score":0.24230000376701355,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T12122","display_name":"Physical Unclonable Functions (PUFs) and Hardware Security","score":0.1703999936580658,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.10610000044107437,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.5479999780654907},{"id":"https://openalex.org/keywords/enhanced-data-rates-for-gsm-evolution","display_name":"Enhanced Data Rates for GSM Evolution","score":0.5212000012397766},{"id":"https://openalex.org/keywords/system-on-a-chip","display_name":"System on a chip","score":0.3490000069141388},{"id":"https://openalex.org/keywords/chip","display_name":"Chip","score":0.2840000092983246},{"id":"https://openalex.org/keywords/domain","display_name":"Domain (mathematical analysis)","score":0.27950000762939453}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7577000260353088},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5479999780654907},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.5212000012397766},{"id":"https://openalex.org/C118021083","wikidata":"https://www.wikidata.org/wiki/Q610398","display_name":"System on a chip","level":2,"score":0.3490000069141388},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3280999958515167},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.32089999318122864},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.29420000314712524},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.289000004529953},{"id":"https://openalex.org/C165005293","wikidata":"https://www.wikidata.org/wiki/Q1074500","display_name":"Chip","level":2,"score":0.2840000092983246},{"id":"https://openalex.org/C36503486","wikidata":"https://www.wikidata.org/wiki/Q11235244","display_name":"Domain (mathematical analysis)","level":2,"score":0.27950000762939453},{"id":"https://openalex.org/C193536780","wikidata":"https://www.wikidata.org/wiki/Q1513153","display_name":"Edge detection","level":4,"score":0.27129998803138733},{"id":"https://openalex.org/C106516650","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm design","level":2,"score":0.2628999948501587},{"id":"https://openalex.org/C46743427","wikidata":"https://www.wikidata.org/wiki/Q1341685","display_name":"Inference engine","level":3,"score":0.2531999945640564}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/lca.2026.3703423","is_oa":false,"landing_page_url":"https://doi.org/10.1109/lca.2026.3703423","pdf_url":null,"source":{"id":"https://openalex.org/S17643076","display_name":"IEEE Computer Architecture Letters","issn_l":"1556-6056","issn":["1556-6056","1556-6064","2473-2575"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Computer Architecture Letters","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"LLM":[0],"inference":[1],"on":[2,24],"unified-memory":[3],"(UM)":[4],"edge":[5],"SoCs":[6],"is":[7,55,139],"commonly":[8],"assumed":[9],"to":[10,47,69],"follow":[11],"a":[12,32,74],"GPU-centric":[13,52],"execution":[14,43,112,137],"model":[15],"once":[16],"nominal":[17,106],"memory":[18,76],"capacity":[19,107],"appears":[20],"sufficient.":[21],"Our":[22],"experiments":[23],"an":[25,126],"NVIDIA":[26],"Jetson":[27],"Orin":[28],"Nano":[29],"8GB":[30],"reveal":[31],"prefill":[33,122],"feasibility":[34,128,138],"boundary":[35,67,80],"in":[36,118],"the":[37,66,119],"evaluated":[38,120],"modified":[39,75],"FlexGen-based":[40],"runtime,":[41],"where":[42],"stability":[44],"becomes":[45,83],"sensitive":[46],"runtime":[48],"memory-management":[49],"behavior.":[50],"Under":[51],"execution,":[53],"collapse":[54,97],"first":[56],"observed":[57],"at":[58],"batch":[59,70],"size":[60,71],"8":[61],"with":[62,85],"zram":[63],"enabled,":[64],"and":[65,131],"shifts":[68],"16":[72],"under":[73],"configuration.":[77],"At":[78],"these":[79],"points,":[81],"completion":[82,123],"probabilistic,":[84],"only":[86,135],"23%\u201333%":[87],"of":[88],"runs":[89],"succeeding;":[90],"all":[91],"failures":[92],"occur":[93,99],"before":[94],"decode.":[95],"Notably,":[96],"can":[98],"despite":[100],"apparent":[101],"physical-memory":[102],"headroom,":[103],"showing":[104],"that":[105],"alone":[108],"does":[109],"not":[110],"determine":[111],"stability.":[113],"These":[114],"findings":[115],"show":[116],"that,":[117],"setup,":[121],"acts":[124],"as":[125],"effective":[127],"gate:":[129],"latency":[130],"throughput":[132],"become":[133],"meaningful":[134],"after":[136],"established.":[140]},"counts_by_year":[],"updated_date":"2026-07-18T05:51:51.687321","created_date":"2026-06-16T00:00:00"}
